Positive audience sentiment appears to be correlated with mentions of jobs(A), and we see another spike toward the end of the speech, corresponding to mentions of both jobs and health care.(B)  The audience seemed to trend toward the negative once near the beginning of the speech.

Above zero denotes a positive sentiment for the blue line, and all other lines represent the volume of mentions over time. The start and ending of the speech are marked.

Twitter Audience Sentiment and Mentions

Sentiment is tracked by counting the number of tweets containing positive and negative keywords from a list. Tweets were restricted to those containing one or more of ‘Obama’, ‘SOTU’, and ‘StateoftheUnion’.

The Python code used can be found on Google Code under TwitterTrends, and the entire data set is available from InfoChimps.